JNK2 Translocates to the Mitochondria and Mediates Cytochrome c Release in PC12 Cells in Response to 6-Hydroxydopamine*

6-Hydroxydopamine (6-OHDA) causes death of dopaminergic neurons by mitochondrial dysfunction with JNKs as central mediators. Here we provide novel insights into specific actions of JNK isoforms in 6-OHDA-induced death of PC12 cells. Twenty five μm 6-OHDA enhanced total JNK activity in the cytoplasm, nucleus, and at the mitochondria. Inhibition of JNKs by 2 μm SP600125 or transfection with dominant-negative JNK2 (dnJNK2) rescued more than 60% of the otherwise dying PC12 cells after 24 h, whereas transfection with dnJNK1 had no protective effects. In contrast to constitutively present JNK1, JNK2 amounts increased in the nucleus and at the mitochondria after 6-OHDA stimulation. JNK inhibition by SP600125 or transfection of dnJNK2 reduced the pool of active JNKs in the nucleus, the release of cytochrome c, as well as the cleavage of caspase-3 and its substrate poly(ADP-ribose) polymerase-1. Transfection with dnJNK1, however, had no effects on the translocation of JNKs to the mitochondria or the release of cytochrome c. Our data provide novel functional insights into the pathological role of individual JNK isoforms, the signalosome at the mitochondria, and the mode of JNK-induced release of cytochrome c.


6-Hydroxydopamine (6-OHDA) causes death of dopaminergic neurons by mitochondrial dysfunction with
JNKs as central mediators. Here we provide novel insights into specific actions of JNK isoforms in 6-OHDAinduced death of PC12 cells. Twenty five M 6-OHDA enhanced total JNK activity in the cytoplasm, nucleus, and at the mitochondria. Inhibition of JNKs by 2 M SP600125 or transfection with dominant-negative JNK2 (dnJNK2) rescued more than 60% of the otherwise dying PC12 cells after 24 h, whereas transfection with dnJNK1 had no protective effects. In contrast to constitutively present JNK1, JNK2 amounts increased in the nucleus and at the mitochondria after 6-OHDA stimulation. JNK inhibition by SP600125 or transfection of dnJNK2 reduced the pool of active JNKs in the nucleus, the release of cytochrome c, as well as the cleavage of caspase-3 and its substrate poly(ADP-ribose) polymerase-1. Transfection with dnJNK1, however, had no effects on the translocation of JNKs to the mitochondria or the release of cytochrome c. Our data provide novel functional insights into the pathological role of individual JNK isoforms, the signalosome at the mitochondria, and the mode of JNK-induced release of cytochrome c.
For the understanding of JNK actions, crucial issues remain to be elucidated, such as the JNK isoform(s) responsible for cytochrome c release or for changes in gene expression. Present concepts grant JNK1 a more physiological and anti-apoptotic role (16,(33)(34)(35), whereas degenerative-apoptotic actions are attributed to JNK3 (36,37). Is the JNK-mediated apoptosis downstream of mitochondrial dysfunction also isoform-specific? This study analyzes the contribution of different JNK isoforms to 6-OHDA-induced death of PC12 cells. Neurotoxicity caused by 6-OHDA is widely used as an experimental model for the neuronal pathology of Parkinson's disease; it involves mitochondrial dysfunction (38) and activation of the MAP kinase kinase (MKK)/JNK/c-Jun pathway (11, 19, 39 -42). Activation of JNKs might be particularly relevant for the pathogenesis of Parkinson's disease. Under experimental conditions, inhibition of MKK4, JNKs, and their nuclear substrate c-Jun has a significant protective effect against the degeneration of dopaminergic neurons in the substantia nigra and the decay of dopaminergic fibers in the striatum (10,13,41). In consequence, the indirect JNK inhibitor CEP-1347, an antagonist of the mixed lineage kinases, currently undergoes phase II/III trial as add-on for the treatment of Parkinson's disease in humans. We used the model system of PC12 cells, because these cells express only JNK1 and JNK2, but not JNK3 (43,44), allowing a subtractive evaluation of the functions of JNK1 and JNK2.
Here we provide evidence that JNK2, but not JNK1, is a central mediator of cytochrome c release following 6-OHDAinduced death of PC12 cells. In contrast to the JNK1 isoform, JNK2 translocates to the nucleus and the mitochondria where it acts downstream of MKK4. Thus, the difference in activation and translocation of JNK isoforms suggests the existence of separate apoptotic JNK signalosomes. These findings have considerable implications for therapeutic strategies targeting JNK signaling in neurodegenerative diseases.

EXPERIMENTAL PROCEDURES
Cell Culture and Treatment of Cells-PC12 cells were cultured as described previously (45). Transfected cell clones were kept in the presence of 500 g/ml G418 to maintain selection. Subconfluent cells were incubated with 25 M and 50 M 6-OHDA, respectively, for the indicated times. One or 2 M SP600125 (in Me 2 SO) were added 30 min before 6-OHDA stimulation, if not indicated otherwise.
Plasmid Construction and Transfection-cDNA of JNK2␣1 (SAPKA/ SAPK ␣I; GenBank TM accession number L27111) from rat brain was mutated by replacing Thr 183 with an alanine and Tyr 185 with a phenylalanine, using PCR site-directed mutagenesis. The resulting dominant-negative (dn)JNK2 was inserted into the expression vector pEGFP-C3 as a HindIII/SalI fragment. Transfection of the dnJNK2-EGFP construct or the EGFP vector was performed using the Trans-Fast transfection system (Promega) according to the manufacturer's instructions. Individual stable clones that expressed the dnJNK2-EGFP construct or EGFP were selected and used for experiments as described recently (46). Dominant-negative JNK1 was kindly provided by Dr. Tuula Kallunki (Danish Research School in Molecular Cancer Research, Copenhagen, Denmark).
Trypan Blue Exclusion Assay-Cell viability was investigated by trypan blue exclusion in combination with cell counting as described previously (45). At least 4 sets of independent experiments were conducted. If not indicated otherwise, cells were counted 24 h post-stimulus. Cell counting was in part performed by investigators blinded to the identity of the samples in order to avoid bias.
Whole Cell Extracts and Nuclear and Cytoplasmic Extracts-Whole cell lysates were generated from subconfluent cells. Before harvesting, cells were washed with phosphate-buffered saline. For whole cell extracts, cells were resuspended in denaturing lysis buffer (20 mM Tris (pH 7.4), 2% SDS, 1% phosphatase inhibitor, and 1% protease inhibitor), incubated at 95°C for 5 min, briefly sonicated, and centrifuged to remove insoluble material (15,000 ϫ g for 15 min). To obtain nuclear and cytoplasmic extracts, cells were lysed as described previously (46). Protein extracts were stored at Ϫ80°C. The protein concentration of the supernatant was determined by the Bio-Rad protein assay, with bovine serum albumin as standard. The purity of the nuclear and cytoplasmic fractions was tested with an antibody directed against activating transcription factor-2 (ATF2).
Preparation of Mitochondria-For mitochondria preparation, cells were washed twice with phosphate-buffered saline and harvested. All the following steps were conducted at 4°C. Cells were washed and resuspended in sucrose buffer (20 mM HEPES, 10 mM KCl, 1.5 mM MgCl 2 , 1 mM EGTA, 1 mM EDTA, 250 mM sucrose, 0.1 mM phenylmethylsulfonyl fluoride, 1 mM dithiothreitol, protease inhibitor), stored for 1 h on ice, and lysed by aspirating cells through a 27-gauge syringe. The lysate was centrifuged at 750 ϫ g, and supernatant was collected and centrifuged at 10,000 ϫ g. The pellet was washed in sucrose buffer and lysed in denaturing lysis buffer (see "Whole Cell Extracts and Nuclear and Cytoplasmic Extracts"). The absence of relevant cytoplasmic contamination was confirmed by measuring the lactate dehydrogenase activity (Roche Applied Science) according to the manufacturer's instructions. The purity of the mitochondrial and cytoplasmic fractions was tested with an antibody directed against GRP75.
RT-PCR-Total RNA was isolated from PC12 cells 18 and 24 h after addition of 25 M 6-OHDA using Trizol (Invitrogen). 0.2 volume of chloroform was added to the Trizol extract, and both were mixed and incubated for 3 min at room temperature. Samples were then centrifuged (15 min; 16,000 ϫ g; 4°C), and the upper phase was transferred to a fresh 1.5-ml tube. The RNA was precipitated with 2-propanol and washed with 75% ethanol. Pellets were air-dried and dissolved in RNase-free (diethyl pyrocarbonate-treated) water. After quantification and quality validation, the remaining RNA was aliquoted and stored at Ϫ80°C.
The RT reaction was performed using the SuperScript First-Strand Synthesis System (Invitrogen) for RT-PCR according to the manufacturer's instructions with 0.5 g of RNA for each sample. The subsequent PCR was carried out with the PCR reagent system (Invitrogen) by using 2 l of cDNA, 2 M MgCl 2 , 0.2 mM of each dNTP, 0.2 M of each primer, and 2.5 units of TaqDNA polymerase. For normalization, the histone H2A variant H2A.z was used. The absence of contamination was checked by negative control assays, in which RNA in the RT reaction or cDNA in the PCR were replaced by diethyl pyrocarbonate-treated water. Amplification was performed with the following primer sets: 1) H2A.z: sense 5Ј-CGTATTCATCGACACCTGAAA-3Ј, and antisense 5Ј-CTGTTGTCCTTTCTTCCCAAT-3Ј; 2) bim: sense 5Ј-CAACACAAAC-CCCAAGTCCT-3Ј, and antisense 5Ј-TCTTCCGCCTCTCGGTAAT-3Ј. The specificity of the primer pairs was verified by comparative alignment using the BLAST data base (NCBI, National Institutes of Health, Bethesda, MD). All amplicons were cloned and sequence-verified.
Statistics-All experiments were carried out independently 3-6 times. Statistical analysis was performed with GraphPrism software using one-way analysis of variance with repeated measures. Means were compared using the post-hoc Bonferroni test, and significance was defined for p Ͻ 0.05. Data were expressed as mean Ϯ S.E.

RESULTS
Activation of JNK and c-Jun-In the initial experiments, we analyzed the effect of 6-OHDA on the distribution and/or activation of JNKs and their nuclear substrate, the transcription factor c-Jun. To test the purity of the nuclear and cytoplasmic fractions, we performed Western blots with an antibody directed against the nuclear protein ATF2 (Fig. 1A). Twenty-five and 50 M 6-OHDA dose-dependently induced a distinct phosphorylation of JNKs in the cytoplasm and the nucleus (Fig. 1B), whereas 10 M 6-OHDA was less effective (Fig. 1C). Concomitantly, the N-terminal phosphorylation of c-Jun dose-dependently increased in the nucleus (Fig. 1B). Compared with other strong JNK stimuli, 6-OHDA evoked a more delayed time course of JNK activation with maximal activation after 4 h in the cytoplasm and nucleus (Fig. 1B). According to this time course, the majority of experiments was performed at the time point of maximal JNK activity, i.e. 4 h after 6-OHDA stimulation (see below).
The concentration of 6-OHDA also affected the distribution of JNK isoforms. Whereas the constitutive presence of JNK1 and JNK2 did not change in the cytoplasm following 6-OHDA, the amount of JNK2 dose-dependently increased in the nucleus between 10 and 50 M 6-OHDA (Fig. 1C). Thus, the augmentation in total phospho-JNK in the nucleus paralleled the increase of JNK2.
Most interestingly, the pattern of the JNK isoforms did not only differ among each other but also between different compartments with a prominent JNK1 band at 46 kDa and a prominent JNK2 band at 54 kDa in the nucleus, whereas the 54-and the 46-kDa bands, respectively, were more pronounced in the cytoplasm (Fig. 1C).
Inhibition of JNK Protects against 6-OHDA-induced Death-For inhibition of JNK activity, the direct and specific JNK inhibitor SP600125 was added to the cell culture medium 30 min prior to 6-OHDA. To demonstrate the SP600125-mediated inhibition of catalytic JNK activity, we determined the phosphorylation of the nuclear JNK substrate c-Jun. Application of 1 or 2 M SP600125 substantially reduced or completely prevented the N-terminal phosphorylation of c-Jun, whereas the phosphorylation of JNKs by upstream kinases was not affected ( Fig. 2A). By application of SP600125 we could demonstrate to what extent JNKs contributed to cell death induced by 6-OHDA. As analyzed in trypan blue exclusion assays, addition of 25 M 6-OHDA caused substantial cell death (36%) after 24 h (Fig. 2B). The preincubation with 1 or 2 M SP600125 rescued 51 and 68% of the otherwise dying PC12 cells (Fig. 2B). The time window for SP600125-mediated protection is smaller than 4 h, because JNK inhibition was still effective when applied 2 h but not 4 h after 6-OHDA stimulation (Fig. 2C). The inhibition of JNKs, however, does not provide a permanent protection. Seventy two hours after 25 M 6-OHDA (without change of the medium), the survival rates did not differ between the SP600125-treated cells and controls (data not shown). Finally, SP600125 could not prevent cell death induced by 50 M 6-OHDA at any time point investigated (Fig. 3).
Protection by Dominant-negative JNK2-The translocation of JNK2 into the nucleus, which paralleled the increase of nuclear JNK activity, raised the question whether JNK2 is responsible for 6-OHDA-mediated death rather than the more "physiological" or even "protective" JNK1 (16,35). For this purpose, PC12 cells were stably transfected with EGFP-tagged dnJNK1 and dnJNK2. Dominant-negative JNK2, but not dnJNK1, protected PC12 cells from cell death induced by 25 M 6-OHDA (Fig. 3). However, similar to SP600125, dnJNK2 did not protect against 50 M 6-OHDA, and its protective effect was also lost after 72 h (see above).
Transfection of dnJNK2 interfered with the activation of JNKs. Most importantly, dnJNK2 reduced the pool of phosphorylated total JNK in the nucleus after stimulation with 6-OHDA to a much higher extent than in the cytoplasm, whereas dnJNK1 had no such effect (Fig. 4A). Vector controls neither affected cell death (Fig. 3) nor the activation of endogenous JNKs (Fig. 4A). Transfection with dnJNK2 strongly reduced the 6-OHDA-induced translocation of endogenous JNK2 to the nucleus in contrast to the findings in wild type (Fig. 1C) or vector-transfected cells (Fig. 4B). The construct itself was almost completely excluded from the nucleus because it was barely detectable in nuclear preparations but was clearly visible in the cytoplasm (Fig. 4B).
Translocation of JNK2, but Not JNK1, to the Mitochondria-Because JNK2 translocated to the nucleus after 6-OHDA treatment, we investigated whether JNK2 also translocates to the mitochondria, the central target organelles of 6-OHDA-mediated pathology. We prepared mitochondrial preparations devoid of cytoplasmic contaminations. To test the purity of our mitochondrial preparations, we performed Western blots with an antibody directed against the mitochondrial heat shock protein GRP75 (Fig. 5A). In untreated PC12 cells, mitochon- drial preparations contained a substantial amount of JNK1, and this pool did not change following 6-OHDA treatment (Fig.  5B). Most interestingly, we observed a translocation of dnJNK1 to the mitochondria after addition of 6-OHDA (Fig. 5B).
In contrast to JNK1, the amount of JNK2 present at the mitochondria increased at the mitochondria between 1 and 2 h and reached its maximal level at 4 h (Fig. 5C). Dominantnegative JNK2 could not be detected at the mitochondria. Moreover, dnJNK2 did not interfere with the increase of endogenous JNK2 at the mitochondria following 25 M 6-OHDA (Fig. 5C) or with the mitochondrial pool of JNK1 (Fig. 5B). A 30-min preincubation with the direct JNK inhibitor SP600125 (2 M) prevented the translocation of JNK2 to the mitochondria but not the constitutive presence of JNK1 (data not shown).
Consequently, we determined the pool of activated total JNK at the mitochondria (Fig. 5D). Under basal conditions, hardly any phosphorylated JNK could be detected. After stimulation with 6-OHDA (25 M), the amount of phosphorylated JNK increased at the mitochondria, reached its maximal levels at 4 h, and returned to basal levels until 48 h after stimulation.
MKK4, MKK7, JIP-1, and MKP7 at the Mitochondria-Activation and distribution of JNKs are closely controlled by upstream kinases and scaffolds of the JNK signalosome (47,48). Therefore, we investigated the presence and/or the stress-induced activation of the MAP kinase kinases MKK4 and MKK7, the MAP kinase phosphatase MKP7, and the scaffold protein JIP-1 in preparations of PC12 cell mitochondria (Fig. 6).
A substantial amount of nonphosphorylated MKK4 was detected in mitochondria from untreated PC12 cells (Fig. 6A). This mitochondrial MKK4 pool was activated within 1 h after a 6-OHDA stimulus, and its activation reached its maximal levels between 2 and 4 h (Fig. 6A). Most interestingly, transfected dnJNK2 abrogated the phosphorylation, but not the presence, of MKK4 at the mitochondria, whereas transfected dnJNK1 had no effect (Fig. 6A). In contrast to MKK4, MKK7 was not detectable in mitochondrial preparations of PC12 cells either in controls or following 6-OHDA but only in cytoplasmic extracts (Fig. 6B). Western blots of the JNK scaffold protein JIP-1 generated a strong signal in mitochondrial preparations that did not change after 6-OHDA stimulation (Fig. 6B), underlin-ing the specificity and selectivity of the observed localization and activation of JNK signalosome components. The presence of only weakly phosphorylated JNK at the mitochondria suggests the presence of mitochondrial phosphatases that de-activate JNKs. Therefore, we investigated the MAP kinase phosphatase 7 (MKP7) which was recently identified as a major and JNK-specific phosphatase (28). Indeed, MKP7 was detectable at the mitochondria and in the cytoplasm; its expression did not change in response to 6-OHDA (Fig. 6B).
SP600125 and dnJNK2, but Not dnJNK1, Prevent the Cytochrome c Release and Apoptosis in Response to 6-OHDA-The release of cytochrome c is indispensable for the formation of the apoptosome, and the generation of this complex is the initiative event for the activation of downstream caspases (28). Therefore, the last set of experiments addressed the question whether JNKs and in particular JNK2 are involved in the release of cytochrome c.
Already 25 M 6-OHDA caused a strong depletion of mitochondrial cytochrome c, and this cytochrome c loss was substantially inhibited by preincubation of the cells with the direct JNK inhibitor SP600125 (2 M; Fig. 7A). Similar to SP600125, transfection with dnJNK2 completely abrogated the release of cytochrome c in response to 25 M 6-OHDA but was hardly effective against 50 M 6-OHDA (Fig. 7B). Dominant-negative JNK1 had no protective effect (Fig. 7B). The pivotal mediator function of JNK2 in apoptosis induction by 6-OHDA was confirmed by Western blot analysis of caspase-3 and PARP-1 cleavage, which could be blocked by dnJNK2 and, to a lesser extent, by a 30-min preincubation with 2 M SP600125 (Fig. 7C). The absence of caspase-8 cleavage after stimulation with 6-OHDA underlined the exclusive function of the mitochondrial pathway in this apoptotic context (Fig.  7C). Taken together, the inhibition of the cytochrome c release by SP600125 or dnJNK2 tightly correlates with their protection of PC12 cells against 25 M, but not 50 M, 6-OHDA, indicating a shift of pathological pathways toward necrosis at higher concentrations of 6-OHDA (28).
Regulation of bim-In order to investigate whether JNK2 regulates bim expression in the context of 6-OHDA-induced mitochondrial apoptosis, we analyzed bim mRNA levels by semi-quantitative RT-PCR. In wild type and vector-transfected cells, no significant induction of bim could be observed at 18 and 24 h after stimulation with 25 M 6-OHDA (Fig. 8). Most interestingly, dnJNK2 markedly reduced bim mRNA after 18 and 24 h (Fig. 8) but not after 4, 6, or 12 h (data not shown) (Fig.  8). These findings suggest that whereas bim is not induced in response to 6-OHDA in PC12 cells, JNK2 still participates in controlling bim transcription, indicating common mechanisms in a cell type-specific process. DISCUSSION The present study investigates novel isoform-specific functions of JNKs in the mitochondrial apoptosis pathway in PC12 cells. The neurotoxin 6-OHDA, a generator of reactive oxygen species (ROS) and activator of mitochondrial stress (49), kills a high proportion of PC12 cells within 24 h. Our data show that JNK2, but not JNK1, is translocated to the nucleus and to mitochondria upon 6-OHDA challenge and mediates 6-OHDAinduced apoptosis. Inhibition of JNK activity by SP600125 or transfection with dnJNK2, but not dnJNK1, substantially attenuated 6-OHDA induced cell death, release of cytochrome c, and caspase-3 activation. The presence of the upstream kinase MKK4, but not MKK7, at the mitochondria suggests a defined composition of the JNK signalosome in the mitochondrial stress pathway.
Neurodegeneration and JNK Activation Induced by 6-OHDA-The neurotoxin 6-OHDA is a widely used experi- mental stimulus for the death of (dopaminergic) neurons in vivo and in vitro. Its various effects of action comprise the generation of ROS with a subsequent imbalance of redox potential regulation, lipid peroxidation, or DNA strand breaks (reviewed in Refs. 50 and 51). Independent of ROS formation, 6-OHDA interacts with and inhibits the complex I in isolated brain mitochondria (52)(53)(54), but the relevance of complex I inhibition for cell death has not yet been proven (55). Tightly related to the mitochondrial pathology is the 6-OHDA-mediated cytochrome c release with subsequent activation of caspases 9 and 3, PARP-1 cleavage, and DNA fragmentation (11,50,51,56,57). Beyond generation of ROS and mitochondrial dysfunction, 6-OHDA is a strong activator of JNKs (11), but the mode of JNK activation by 6-OHDA remains to be elucidated.
We found a protracted activation of JNKs with an increase from 2 h and increasing to maximal levels at 4 h in the cytoplasm, nucleus, and at the mitochondria. This activation was accompanied by a translocation of JNK2 into the nucleus and to the mitochondria, whereas the constitutive presence of JNK1 did not change in these compartments. The differential translocation of JNK isoforms is a critical element for the understanding of isoform-specific JNK actions. For instance, the nuclear translocation of activated JNK2 and JNK3, but not JNK1, is essential for the c-Jun/AP-1 mediated transcription (6). It is also conceivable that the JNK2 translocation into the nucleus leading to an increased nuclear JNK activity and the subsequent c-Jun phosphorylation enhances the expression of Fas ligand, which in turn facilitates apoptosis (58). In contrast, the initiation of mitochondrial apoptosis by JNKs is independent of transcriptional effects (59). Therefore, the transcriptional outcome of JNK signaling has to be differentiated from other strictly localized protein interactions in which JNKs participate. In the mitochondrial context, JNKs are not only involved in the release of cytochrome c, but also in the release of Smac (60). Furthermore, JNKs phosphorylate BimEL (17,61) and Bcl-2 directly at the mitochondria (62). In our study, the amount of translocated JNK2 corresponds to the increase in compartment-specific activation of total JNK, suggesting that JNK2 is the active JNK isoform that mediates the neurodegenerative effects of 6-OHDA.
JNKs Are Mediators of 6-OHDA-induced Cell Death-Inhibition of JNK activity by 2 M SP600125 rescues 68% of otherwise dying neurons, and a similar effect is reached by dnJNK2. JNK-mediated neuronal death occurs in response to various stimuli, e.g. growth factor withdrawal (63), excitotoxicity (36), or neurotoxins such as 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) (41,64). The rescue by dnJNK2 demonstrates that one individual JNK isoform rather than the total pool of JNKs is responsible for cell death as already shown for JNK3 (8). In another model of dopaminergic degeneration, which also implies formation of ROS and mitochondrial dysfunction, JNKs act as degenerative players downstream of MPTP (12,41). In animals, inhibition of JNKs by inactive MKK4 or increased JIP-1 confers powerful protection against MPTP (12,41). Recently, the role of JNKs for 6-OHDA-mediated death was examined with regard to cell death caused by ER stress (65) in response to 40 -100 M 6-OHDA. At higher concentrations, 6-OHDA induces a different, i.e. accelerated (necrotic) cell death (66). The overload of mitochondrial calcium during excitotoxicity can result in spillover of calcium to the ER with subsequent ER stress-mediated death (67,68).
JNK2 Controls the Release of Cytochrome c and the Expression of bim-Cytochrome c released from the mitochondrial transition pore is indispensable for the generation of the apo- ptosome with the subsequent activation of downstream caspases (reviewed in Ref. 69). Therefore, the control of cytochrome c release is central to anti-apoptotic and apoptotic cellular pathways (reviewed in Ref. 42). JNKs can activate pro-apoptotic mediators of the transition pore such as Bax, Bad, Bim, Bid, or Dp5 (17,(25)(26)(27)70) and de-activate antiapoptotic proteins such as Bcl-2 (17,21,29). Moreover, JNKs mobilize Smac/Diablo (30), which is released from mitochondria and inhibits anti-apoptotic proteins in the cytoplasm. The release of cytochrome c is tightly linked to the presence and activation of JNKs (25,71), and the JNK-mediated cytochrome c release can involve the nuclear activation of c-Jun (4) with the subsequent expression of the pro-apoptotic BH3-only protein Bim (4,26). The JNK-mediated cytochrome c release includes some particularities; it is insensitive to cyclosporin A and independent of the permeability transition of the inner mitochondria membrane (30).
The issue of the JNK isoform(s) responsible for cytochrome c release in neuronal cells has not been clarified. Here we show that dnJNK2 antagonizes 6-OHDA-induced cell death, cytochrome c release, and caspase-3 activation to a similar, or even stronger, extent as the JNK inhibitor SP600125. The translocation of JNK2 to the nucleus and mitochondria supports the notion that the assembly of a selective JNK2 signalosome propagates the mitochondrial pathology. A central role of JNK2, but not JNK1, for cellular degeneration was also seen in fibroblasts where JNK2 mediates tumor necrosis factor-␣-induced cell death via caspase and cathepsin signaling (72). Most importantly, SP600125, which prevents the stress-induced alterations in the membrane potential (35), blocks the mitochondrial translocation of JNK2 but not JNK1 (data not shown). This finding suggests that intracellular distribution of JNK2 depends on its activation and, by a positive feedback mechanism, on activated upstream kinases (73) and on the availability of JNK2 that can be activated. It remains to be clarified whether the lacking mobility of JNK1 might be due to the absence of this feedback loop.
The induction of apoptosis and the activation of caspase-3 by 6-OHDA has been examined in several studies (74,75). Especially in PC12 cells, it has been shown that 25 M 6-OHDA caused apoptosis, whereas the application of 50 M 6-OHDA caused necrotic cell death (66,76). We did not detect an activation of caspase-8 after addition of 25 M 6-OHDA, which underlines the importance of the mitochondrial pathway. In fact, the 6-OHDA-induced activation of caspase-8 seems to be a cell type-specific event, because it has only been described to occur in MN9␤ cells and mesencephalic neurons (77). The crucial involvement of JNKs in the mitochondrial stress pathway after stimulation with 6-OHDA and its effect on the activation of caspase-3 and the cleavage of PARP-1 has not been examined before. However, it is known that JNKs play a role in mitochondrial signaling in the context of oxidative stress, where JNK inhibition reduced caspase-9 activity (9).
In addition to the phosphorylation of mitochondria-associated substrates, nuclear JNK2 adds to the nuclear pool of activated JNKs and subsequently contributes to the transcription of c-Jun/AP-1 controlled target genes such as bim (4,17), and SP600125 or dnJNK2 distinctly reduce nuclear levels of activated JNKs or JNK2, respectively. Dominant-negative JNK2 interfered with the transcription of bim at a delay between 18 and 24 h but not between 4 and 12 h. The role of the constitutively present JNK1 for mitochondrial functions remains to be defined. JNK1 was recently found in the antiapoptotic XIAP-TAK1 cascade (35), but the link to mitochondria-associated protection is speculative. Moreover, the clear distinction between JNK1 and JNK2 is not a general principle because JNK1 can also confer apoptosis by Bcl-2 phosphorylation in immune cells (78).
The JNK Signalosome at the Mitochondria-The specific mitochondrial translocation of JNK2, but not JNK1, raises the question which signaling pathways control the JNK translocation to and/or the JNK activation at the mitochondria. In leukemia cells, it has been shown that phosphorylation of JNKs is a prerequisite for translocation, which is supported by the absence of dnJNK2 at the mitochondria. On the other hand, the presence of nonphosphorylated JNK1 and JNK2 in untreated cells argues against phosphorylation as an absolute precondition for translocation. Similarly, the inhibition of MKK4 phosphorylation by dnJNK2 does not reduce the amount of MKK4 at the mitochondria, indicating that translocation and phosphorylation can be separate events. The unexpected inhibitory effect of dnJNK2 on MKK4 phosphorylation is a novel observation that warrants further investigation. It is well conceivable, and supported by our data so far, that a constitutive pool of MKK4 is present at the mitochondria, part of which is activated or replaced with activated MKK4 from the cytoplasmic environment upon 6-OHDA stimulation. Overexpression of dnJNK2 could shift the equilibrium between active and inactive MKK4 by recruiting an excessive amount of active MKK4 from the mitochondria-associated pool after 6-OHDA stimulation, which may result in the observed absence of the mitochondrial phospho-MKK4 signal in dnJNK2-transfected cells.
We have identified MKK4 and JIP-1 to be localized at the mitochondria in our system, whereas MKK7 was not detectable in the mitochondrial fraction. This suggests a specific JNK pathway for mitochondrial pathology. MKK4 was found previously in mitochondrial fractions from cardiomyocytes (23), whereas the presence of JIP-1 and the absence of MKK7 are completely novel observations. Furthermore, the co-localization of activated JNK with the JNK-binding protein Sab (79) points to specific mitochondrial components of the JNK pathway. Sab strongly associated with the smaller JNK splice variants (79), which is in agreement with our finding that activated JNK was almost exclusively found in the 46-kDa band (Fig. 5C).
JNK-mediated Cell Death, the Issue of the Context-Inhibition of JNKs only provides a transient protection against 25 M 6-OHDA because SP600125 and dnJNK2 are no longer protective after 72 h. These data were confirmed by our recent findings that the inactivation of the N-terminal phosphorylation domains of c-Jun (80) conferred significant but only transient protection against 6-OHDA-mediated degeneration of dopaminergic neurons (81). Transient protection against 6-OHDA was also reported after inhibition of caspases (56). This critical issue of temporary protection is widely neglected, and in consequence, false-positive assumptions of protective strategies have been made in the literature.
Besides temporal restriction, neither SP600125 nor dnJN2 protected against 50 M 6-OHDA, indicating the shift from JNK-dependent to JNK-independent cell death, e.g. by ER stress (66). This transition between apoptotic and necrotic cell death deserves close attention. Neurodegeneration is a multifactorial process demanding different defense strategies. The shift of "necrotic" signal pathways to "apoptotic" signal pathways does not yet rescue the damaged neuron but does open the chance for additional preservative strategies and widens the time frame of therapeutic intervention, e.g. by selective inhibition of JNK isoforms. In summary, the finding that specific and exclusive JNK2 signalosomes control the mitochondrial apoptosis pathway in the model system of 6-OHDA-stimulated PC12 cells may contribute to a future catalog of context-specific intervention strategies in the treatment of neuronal disorders.